Centrifugal Pump Startup: Suction Voiding

[vpl]

I'm investigating an issue where suction voiding during the pump startup may be an issue and I'm trying to get more of a feel about what's going on.

Starting off I have a large (1200 psi, 3300 gpm) turbine-driven pump. The contractor I'm dealing with believes that the pump has a short-term suction void that occurs nearly instantaneously with pump startup (on the order of 0.1 to 0.2 seconds following startup) and then collapses almost as fast as it forms. The contactor says this is occurring because the pump is "pushing" the water out faster than the gravity fed inlet can supply it and because the contractor believes there is an upstream void for the water to fill.

I'm having a hard time understanding, because at the time this void is supposedly occurring, I don't believe the pump is really pumping anything. Also there's 20+ feet of water on the suction side so there is more than adequate NPSH for the pump (8 feet's the low value, and the tanks is being kept well above that.) I've spent three days with the contractor repeating the same information over and over, and I'm getting frustrated. So...

Can anybody shed some light on this for me? Any good sources that I can read to get a better understanding of suction side voiding during pump starts? Is this a feature that will only occur if I have a void somewhere on the discharge side? Are there any sort of bounding values for times when this would occur, how long it would last, and the sort of pressure spike I would get when it collapses?

All help much appreciated.

Patricia Lougheed

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[MikeHalloran]

What sort of pressure spike _did_ you get?

...

Nobody _measured_ anything, right?

;---

Once a centrifugal pump has reached a stable speed, and its system has bled out all the air that's going to come out and the system flow has stopped surging, you can get a good estimate of the flow rate, given the pressure difference, the pump curves, and the rotational speed.

In your case, you're worrying about startup transients in what must be a multistage pump. I assume the pump is not now instrumented to measure the pressures, and there's no tachometer on the shaft, either. Perhaps there once were pressure gages, but they destroyed themselves in some prior startup...

;--

Centrifugal pumps are kinetic devices. If there's water in it, and it's rotating, it's pumping ... something. But it might be more accurate to say that they're doubly kinetic, because they're slinging out water in order to create enough of a pressure difference to induce more water to arrive for them to sling out... i.e., startup is guaranteed erratic.

Just the other day, I started up my lawn sprinkler pump for the season. There was enough water in the volute to bootstrap the process, but for a good two minutes it just sort of burped and spit, and it was pretty wobbly for a good five minutes after that.

Your pump could water my lawn in seconds, if you could get it started without ... without ...

... uh, just what is the actual symptom that caused you and the contractor to begin trading hypotheses?




Mike Halloran
Pembroke Pines, FL, USA

 

[vpl]

well what started it all off was discovery of a void at the high point of the system... right before a closed injection valve. I'm trying to analyze whether there will be a water hammer when the valve opens. The only data available is from routine surveillance tests which rely on a test return line being open and from one special test, also done with the test return line open.

Contractor is saying the size of the void doesn't matter because of the instantaneous suction void when the pump starts. Relying on strip charts of routine pump tests (with the test return line open), they've pointed out that there is the same characteristic on all of them. Pump starts, about 0.10 seconds later there is a small pressure rise, followed by a smaller pressure drop about 0.05 seconds later. After another 0.05 seconds, there is a "spike" to somewhere around 400 psi, as measured on the pump discharge pressure indicator. Over the next 0.30 seconds, the pressure continues to rise, then dips, rises, dips and steadies out -- about the time, the flow indicator is starting to register flow. Sometimes the small drop is basically invisible on the chart. Sometimes, there clearly is not a drop in that first 0.15 seconds. Sometimes there are two or three little spikes between 400 and 450 psi (and in that first 0.20 seconds). Oh, and on the special test, there was a pressure indicator attached to a vent line up on the high point; it registered 1800 psi about 0.60 seconds into the pump start. The contractor says that spike should be ignored as it's due to the pressure being attenuated in the 3/4" line versus the 12" pipe.

While I'd like to believe all this, I need to be sure I wouldn't get a water hammer on a true injection -- with the test return lines closed. I recently attended an ASME short course, and my notes indicate that it's quite possible to have a very short lived suction void right as the pump starts, for the very reasons you mentioned. However, I don't have any information as to how long is "very short" and how big a pressure rise would be reasonable on collapse of that suction void. I also have an "unreasonable" dislike of ignoring data or differences between the test lineup and the real event.

I recognize that I'm not enough of a pump expert to challenge the contractor's conclusion, but I have a gut feel that they're wrong. So that's why the post...

Patricia Lougheed

Please see FAQ731-376 for tips on how to make the best use of the Eng-Tips Forums.

 

[MikeHalloran]

The "special test" comprises ... what? Running the pump into a closed discharge indefinitely after it has started?

This is clearly a nontrivial system. Whoever designed it and sized the pump probably had/has a lot to say about starting it up. You just have to find that person, or whatever record they left for you to find.

How do you infer the existence of a 'void'?



Mike Halloran
Pembroke Pines, FL, USA

 

[UmeshMathur]

Are you sure the pressure spikes are real? Measurement devices have their own idiosyncrasies, and there can be dynamic delays and signal attenuation in the instrument itself, unless it is designed for very high-speed response and data transmission.

Since the suction side is full of liquid and you have no NPSH problems, the suction void theory sounds fishy to me unless you have a lot of dissolved gas in the liquid being pumped (hard to imagine with water).

 

[Artisi]

Just a couple of basic thoughts initially;
1. Is the discharge side fully charged on startup - or do you have to fill or partially fill the "line".
2. Is the inlet side and pump casing completely devoid of air prior to startup, ie, no air-pockets in the pump case/ings etc.
3. You say the pump is turbine driven - can the pump be brought up to speed slowly
4. what is the time lag from start to full speed?


Naresuan University
Phitsanulok
Thailand

 

[25362]

This is a quote from S. Yedidiah's Centrifugal Pump Users' Guidebook Problems and solutions, Chapman and Hall, ISBN 0-412-99111-X, on "hysteresis and other forms of instability" that may apply to the case in hand:

...Recirculation always acts towards increasing the developed head. Consequently when we start out with a closed valve and open it gradually, recirculation eventually stops at a certain flow rate. At that flow rate, the head drops suddenly owing to the termination of recirculation.
............................................................................................
............................................................................................
At present little is known about how recirculation produces a jump in the NPSH requirements of the pump. It seems to be related to the fact that the action of the impeller on the surrounding liquid is of a cyclical nature, caused by the finite number of impeller blades. Occasionally, the frequency of such an action may produce resonance with another cyclic effect occurring in the pump, resulting in a jump.

This excerpt pertains to chapter 9: Recirculation in Part II. PERFORMANCE FACTORS.

 

[RimBone]

Patricia,
There is a company in Seattle (ENSR 425-881-7700) that made a scaled model for one of our pump stations. The cost was around $200k. The recommended solutions proved to be effective. In my case it had multiple problems. Wetwell vortexing, cavitation in the pump suction 90deg elbows, and wetwell water movement; which caused starvation at the pump inlet depending on the location of the pump inlet in the wetwell. These were visible when dye was apply into the model. Instruments measured pressures and flows,

equipment: 200, 400, and 500hp two stage pumps

The head curve is relatively flat, the NPSHA does not meet the NPSHR, inlet conditions (size) were too small, inlet condition created vortexing in the wet well which would come and go, (straight suction pipes into the wet well).

Not all conditions could be corrected, IE the NPSHA, straight suction pipes (no 90 deg suction bells).

Impellers were trimmed so the pump curves would move left closer to the POR. The suction pipe sizes were increased, full flow suction valves were installed, baffles were installed in the wet well to correct the water movement.

I can be reached at 858-292-6482 if you think any of this is applicable to your situation.
thanks
Rim Stanley
City of San Diego
MWWD